effective altruism on the verge of bankruptcy
the philosophical conjuring tricks were more important than the financial ones
sbf was arrested yesterday. this story has captivated the tech and finance worlds for over a month, but not—i will insist—because of the fraud, embezzlement, or dazzlingly-lax accounting. no, it is not good-old-fashioned human vice that makes this story interesting. what makes the story interesting are the wild beliefs that seem to have gotten the whole thing off the ground in the first place.
the story interests us because—even if we can’t all articulate it—we suspect there’s something off about the arguments for crypto, effective altruism, and longtermism. important legal work must be done to understand the financial and accounting deceit in the ftx story, but ultimately what makes this story movie material will be the philosophical motives behind the primary players. i want to propose that we will not be satisfied with an explanation of what happened here until we understand three core philosophical confusions (“conjuring tricks,” after wittgenstein) that propel the central characters forward: first, there is a confusion about money; second, there is a confusion about value; and third, there is a confusion about the future.
financial conjuring tricks: yield farming magic beans
there are three basic ways that to make money on one’s crypto holdings—one is to sell them for more than you bought them, another is to sit on them and earn small rewards for contributing compute to the network, the other is to earn larger rewards by lending one’s holdings at interest to exchanges or crypto banks that promise attractive yield. when performed by centralized institutions, techniques of the last type are called ‘yield farming.’
under the right circumstances crypto yield farming can look like easy money. sbf described these circumstances to tracy alloway, joe weisenthal, and matt levine on an infamous episode of the odd lots podcast in april, 2022. the full transcript is worth reading just for shock value, but the basic steps are this:
make some tokens out of thin air. ideally the tokens should purport to give token-holders some stake in a project of economic or political import.
ignite some demand for the tokens by offering them to investors at attractive (usually double-digit) yields—paid in the token you made up.
mark to market the tokens you made up and held on your balance sheet to inflate it for this exact purpose.
use this inflated balance sheet to collateralize loans for risky bets that could theoretically pay yield back to investors in the form of profits but that probably won’t.
pray for reliably profitable bets and no runs on the liquidity pool.
alameda and ftx appear to have done something like this yield farming with their tokens ftt and serum. a lot of other failed cyrpto exchanges seem to have been up to something similar. when alameda’s bets did not pay off, they appear to have paid back their lenders using capital from ftx liquidity providers.
the financial conjuring tricks in play here should be clear—the made-up tokens really are not worth what they appear to be worth. there are other financial conjuring tricks that ftx engaged in, including possible market manipulation. matt levine’s reporting on the subject is comprehensive and entertaining enough to suffice if one’s interest is mainly economic and legal. to understand the ideology, we have to turn to philosophy.
philosophical conjuring tricks: money, value, and the future
despite his cynical sounding comments about crypto on the black-box oddlots podcast and his recent suggestion that his e.a. and longtermism was just “this dumb game we woke westerners play where we say all the right shiboleths and so everyone likes us,” i think we have a hard time explaining sbf’s behavior unless we understand him to have been genuinely motivated by philosophical beliefs about money, value, and the future.
without this, it’s hard to make sense of his desire to work on animal welfare before will macaskill convinced him to go into finance unless we take him to have persistent altruistic motives. it’s hard to make sense of his public defense of crypto projects beyond what would be needed to sustain ftx unless we take him as a genuine true believer in decentralized monetary regimes. and it’s hard to make sense of the scale of his charitable contributions to longtermist causes unless we see him as genuinely believing they deserved his money.
we take these to be genuinely motivated behaviors; not cynical window-dressing. that is why we cannot resist the story. these beliefs are really strange, but we struggle to articulate exactly why. our collective delight in the story is an expression of ideological vindication.
money
to see the ideological incoherence at the heart of the project, start with the beliefs about money. at its core, crypto is a claim about the kind of thing that money is, and how we ought to arrange our institutions given that fact.
crypto people tend to believe that money is a kind of objective thing whose value is dependent on others coming to accept it as valuable. monetary theorists call this the commodity theory of money. when we tell our children that money originated out of the need to escape impossibly-complex barter systems, we are asserting the commodity theory of money. when crypto people call sovereign cash fiat—to remind us that it’s value is ‘made up’ and dependent on our ‘collective faith’—they are asserting the commodity theory.
in contrast, the credit theory of money claims that money is a kind of promise, between a creditor and a debtor. on the credit theory of money, money is just a credit receipt—something that borrowers and lenders strike to represent the debtors promise to repay a debt and the lenders promise to accept that payment as fulfillment of that debt.
the commodity theory of money faces a number of challenges. first, as david graeber details in his book, debt, the first 5000 years, there is no actual historical evidence that monetary systems ever arose out of barter systems. they all seem to have originated out of complex accounting systems for tracking what people owed each other—especially where what people owed one another was their labor.
second, the commodity theory has a very hard time explaining how it bootstraps the kind of sustainable, non-volatile value needed to constitute a monetary system. the credit theory has no such difficulty because it just locates its value in already-established forms of human debt accounting: an eye for an eye, a tooth for a tooth etc.
the relationship between political power and monetary regime dominance is also neatly predicted by the credit theory, whereas the commodity theory would predict that the best monies would be those that are (ultimately) most technically efficient. credit theories predict that dominant money will track dominant credit, and dominant credit tracks more fundamental forms of dominance (political). throughout recorded history, the most widely-shared credit/debt relation has been that of taxation. the credit theory predicts that the things used to fulfill tax debts will probably be money, and the world’s most reliable creditor will probably be the world’s reserve currency provider. lo and behold, when we look around today that is exactly what we find. if we think it is mysterious why technically inefficient monetary regimes should persist in the face of increasingly-more-efficient private alternatives, we are probably in the grip of the commodity theory. soft sovereign money will always beat out hard private money.
but the entire crypto project basically assumes there is no plausible alternative to the commodity theory. this is wrongheaded. while the credit theory may have some lingering issues, it is the most plausible theory of money we have. money is not a thing with no value independent of what market participants assign to it for expediency. money is an extension of complex and pre-historic networks of debt accounting where value is not assigned by expediency but derived from commonly-recognized debts.
value
sam was die-hard effective altruist. but he appears to have been a motivated altruist before he was convinced that altruism is best realized through a commitment practicing expected value theory. that his altruism predates any of his fame or wealth-seeking seems like prima facie evidence for us to interpret it as persistent and honest motivation. the turning point towards a more effective altruism occurred when moral philosopher will macaskill convinced sam to make as much money as possible so he could maximize the impact of his charitable contributions. will and sam later came to believe (unclear who convinced the other) that very large expected future values could justify some pretty enormous present-day risk (and harm).
after this critical conversation, the groundwork for the financial conjuring trick appeared to unfold fairly naturally. sam went on to trade at jane street, then moved to the bay area where macaskill got him a job working at an effective altruism foundation. sam drank deeply from the e.a. kool aid and eventually went on to start arbitraging bitcoin and then risk-neutrally maximizing expected value through the entities that are unraveling today.
value has an odd place in effective altruism. when will macaskill writes about the project, he takes care to articulate it in a way which attempts to be value neutral—that is, neutral over a particular conception of what is actually the value one is supposed to be effectively bringing about. macaskill and others take care to articulate the view in such a way that maximizing value does not come at the benefit of various “common sense moral constraints.” it is of course, at this point that the view risks triviality. after all, articulating what the common sense moral constraints are and why is just what value theories are about in the first place. to say, “we should do as much of whatever the correct moral theory tells us to do” is probably right, but trivially so.
at this point, macaskill and others go in for a more substantive conception of value because they recognize that their altruism has to be effective. setting aside so-called ‘objective list’ views on what constitutes well-being (the views that are probably right!), macaskill insists that effective altruism should maximize either preference satisfaction or pleasure. that’s because expected value calculations have to be mathematically tractable. they would not be unless you could literally multiply numerical variables to get expected value. this is why effective altruism so often appeals to the mathematically-inclined. if you are disposed to think of life the way you think about computer programs (as consisting, basically, in different sets of optimization problems) you may be drawn to the view.
the challenge though, is that as soon as the view becomes substantive in this way, it opens itself up to the absurd conclusions faced by any moral theory. it is just not at all obvious that maximizing preference satisfaction or pleasure is the right thing to do, even while respecting commonsense moral constraints. one can of course, ceteris paribus one’s way back into triviality, but when in the real world things are hardly ever equal it is unclear how the view actually can provide moral guidance that is simultaneously substantive and non-moronic. in this way, it is much like other theories of value which attempt to circumvent the hard work of developing practical wisdom through intellectual cheat codes. it is not the right way to do ethics.
there are other criticisms of effective altruism—that it launders bad behavior by billionaires hungry for post-facto justifications of their wealth while the california ideology that used to provide this justification crumbles around them. or, that it pacifies global development problems enough to reduce the urgency of government action and the possibility of structural change. or, that it encourages earn-to-give complicity in political-economic careers in institutions that are unjust and exploitative.
these are all important criticisms, but they target consequences of the project from which its proponents could idealize away. criticizing the view’s assumptions about value cut deeper. it is not clear that the effective altruist has an answer to the conceptual issues at the heart of their project.
the future
the longtermist views that seem to have motivated sbf fall out of the effective altruist project fairly naturally. in his recent (and unbelievably popular) book, macaskill argues that if we accept three the following three premises (and a whole lot of embedded ones about value conceived of in the e.a. way), we end up in a surprising place. those three premises are:
future people count
there could be a lot of them
we can make their lives better
these are supposed to be uncontroversial. but as wittgenstein taught, the philosophical conjuring trick usually occurs when we push the language in some innocent-sounding premise past the point at which it actually makes sense. i would insist that this is actually the case for the first premise and the entire philosophical discipline of population ethics that it has unfortunately spawned.
what do we think we mean when we say that future people count? we think we mean that we ought to consider them in our action and planning. they aren’t around now, but they will be around someday, and when they do we don’t want them to suffer because of our actions.
put this way, the premise is as laudable as it is platitudinous. but it gets a substantively different interpretation by a lot of effective altruists who fall under the sway of population ethics. population ethics asks us to compare different possible futures with sets of populations who realize different levels of well-being and then ask which future is preferable. for example, would a world with trillions of people living barely mediocre lives be better than a future world with a billion people living great lives? the so-called “repugnant conclusion” in population ethics appears to be ineradicable: for any finite population with high well being, it is always possible to imagine another world where everyone has lives just barely worth living but where the population is large enough that the effective altruist must prefer it over the higher per-capita welfare world because of its higher aggregate welfare.
if you think these ideas are crazy, outlandish, and could never seriously grip anyone besides a philosopher, you should listen to sbf tell tyler cowen why he would consistently take double-or-nothing bets on the human population (holding well-being constant). he seriously believed that even a terrible action like possibly destroying the human species could be justified if it were just barely outweighed by the possibility that the human species could double in size while maintaining the same or better level of wellbeing. after that sinks in it becomes far less mysterious why he would take such significant and potentially harmful gambles in order to maximize the value of his potential charitable contributions. the possibility of staggeringly-many humans in the distant future just will tend to outweigh many present day moral concerns for simple mathematical reasons: very large numbers crowd out comparatively small ones.
there are reasons for thinking that particular future people do not actually count. after all, they do not exist and might not exist. we generally don’t think of ourselves as owing things to people who do not exist. we can easily derive the idea that we owe things to future generations because we owe things to the current ones, and part of what we owe is that they can feel hopeful, free, and morally unencumbered in their family planning.
if you think that particular future people literally count and that their well being is a function of their preference satisfaction or pleasure this double-or-nothing madness is a conceivable place to end up. it is tragically foolish—like everything else in sam’s story. but it is at least intelligible. that’s the point of understanding the role that these ideological beliefs played in his rise and fall.
***
thus, sam. the picture of a person who probably was not outright vicious; who seems genuinely remorseful; who really did want to make things better but who was in the grip of some very powerful and very bad ideas. perhaps the story shows us the need for more commonsense moral constraints—in the allocation of venture capital, in reporting on crypto, in securities regulation, and in the public reception of wealthy innovators. we have set our future hopes on tech entrepreneurs for too long. the california ideology has just begun to unravel, and it will be more painful. we urgently need another vision of hope and dynamism in america to take its place. it is not yet clear from where such a vision will come.